Normal aging is accompanied by physical decline in brain volume and connectivity in the prefrontal cortex (PFC), along with changes in cognitive abilities related to working memory and executive control. Functions tied to sub-cortical brain structures, such as implicit learning, see relatively less decline in non-pathological aging. In response to declining PFC function, older adults appear to recruit PFC resources differently than young adults as task demands increase. Although much evidence supports the notion that this increased activation is compensatory, inconsistencies and alternative accounts exist. Cabeza & Dennis (2012) recently proposed criteria to clarify conditions under which increased activity should be considered compensatory. The criteria distinguish between successful and unsuccessful compensation and should bring order to the literature on age-related compensatory activation. The criteria, however, do not predict when task conditions should result in compensation. Our understanding of compensatory effectiveness would benefit from systematic examination within a theoretical framework that predicts when task performance should and should not be linked to PFC function. Such a framework exists in the categorization research literature. The theory, called COVIS (for COmpetition between Verbal and Implicit Systems) posits that explicit (verbalizable) category rule learning is mediated by PFC, while implicit (nonverbalizable) rule learning is mediated by subcortical structures. This prediction, when combined with the criteria described above, provides a unique opportunity to increase understanding of compensatory success and failure across a major subdivision of cognitive skill - explicit and implicit reasoning. In the proposed experiments, young and older adults will perform simple category learning tasks that are mediated by either PFC or subcortical regions. PFC blood flow will be measured using functional Near Infrared spectroscopy (fNIR) to measure the timing and duration of changes in activation related to task demands. For explicit rules, compensatory activation should aid performance during learning. During learning of implicit rules, PFC compensation should be attempted but unsuccessful, producing an extended period of increased activity - a hallmark of ineffective compensation predicted by Cabeza & Dennis's criteria. We plan to explore this fundamental prediction from several perspectives, comparing young and older adults, levels of task difficulty, dual task demands, and individual-difference predictors related to brain integrity. This approach holds value for predicting normal age-related brain changes, as well as the results of traumatic injury in different brain regions.